Distributed water-filling algorithm for direct-sequence ultra wideband cognitive radio network with limit on aggregate power emission

Despite their near-noise power emission, ultra wideband (UWB) radios, particularly direct-sequence ultra wideband (DS-UWB), are still possible to cause harmful interference to legacy radio systems. When they participate in a large dense cognitive radio network (CRN) and transmit simultaneously, their joint power emission would be disastrous if without control. We address this problem by imposing a limit on the aggregate power emission of DS-UWB CRN. The limit is negotiable. It takes effect by letting each transmitter adapt to a spectrum void defined as the area below the limit itself and above the interference temperature experienced instantly by the transmitter. We propose a water-filling algorithm that maximises the sum capacity while enabling each transmitter to fit its power spectral density into, and thus to make the most of, the spectrum void. The algorithm is performed locally at the transmitter with low complexity. Numerical analysis based on a realistic office network shows that the algorithm can bring the aggregate power emission even down below the Federal Communications Commission limit for individual UWB device, but still guarantees a given bit-error performance to nearly 40 active users.